Sonder
Interactive Environment,
VR built with Unreal Engine
Sonder, according to the Dictionary of Obscure Sorrow, is the realisation that each random passerby is living a life as vivid and complex as your own. This project, inspired by the Isolation Chamber in the film series Snowpiercer where it is used for therapy sessions, aims to create an environment where safe story telling can happen. In encouraging listening through VR, it also gives purpose to these stories.
Conceptual Foundations
During this period of exploring interactive spaces designed for slowing down and intentional presence, I was deeply influenced by the television series Snowpiercer, particularly Miss Audrey’s Night Car. Within the narrative, the Night Car was initially intended to function as a brothel, but Miss Audrey recognised the need for something else entirely. Through her intervention, part of the Night Car was transformed into the Isolation Chamber. A room designed around brief sessions of isolation and sensory deprivation, where passengers could process grief and loss following the collapse of the world they once knew.
Screen Capture from https://screenrant.com/snowpiercer-audrey-night-car-therapy-isolation-chamber/
This logic of intentional reduction became central to Sonder. This project withholds visual context around who each story belongs to in order to create space for reflection and connection.
This approach was further reinforced through my experience of Laurie Anderson and Hsin Chien Huang’s The Chalkroom, a VR installation that invites viewers to drift through fragments of memory and language. Like the Isolation Chamber, Chalkroom relies on quiet immersion rather than spectacle, demonstrating how reduced sensory input can heighten emotional resonance.
Screen Capture from https://laurieanderson.com/?portfolio=chalkroom
Experiencing The Chalkroom by Laurie Anderson and Hsin Chien Huang at The Singapore Biennale 2019
Immersive Space & Unreal Experiments
Early tests focused on learning Unreal’s spatial and lighting capabilities, using free Marketplace assets to rapidly prototype environments rather than bespoke modelling.
I experimented with Unreal’s water plugin and floating actors, initially imagining a buoyant, weightless environment. However, the resulting motion felt too aquatic. Conceptually, I was aiming for a helium-like suspension rather than water-based physics. This mismatch revealed an early tension between technical affordances and artistic intent.
Environment building play test
Introducing various lighting elements
To cultivate a meditative mood aligned with the project’s ‘zen’ aspiration, I introduced murky blue tones, light shafts, and firefly particles, trying to focus on the interplay of light.
Projection Challenges
The project initially took the form of an immersive room using Unreal Engine supported by three virtual cameras.
Immersive room sketch, which would stretch over 3 walls
Attempting to scale the Unreal environment across three projectors introduced significant technical constraints. Tools like Spout and Syphon were incompatible with my specific engine version, which I was required to use for other version-specific plugins, while professional alternatives like LightAct remained inaccessible. Although Unreal’s nDisplay system worked in principle, configuration issues prevented a successful projection-room deployment. This moment marked the first major pivot. Rather than forcing a projection-based solution, I began considering other forms of immersion.
Interaction Logic with Voice
Initial sketch, imagining how voice could contribute to the environment
A key interaction idea initially was that spoken voice would directly influence the virtual environment. Using TouchDesigner connected to Unreal via OSC, I experimented with how voice could spawn or manipulate geometry in Unreal.
Unreal and TouchDesigner connectivity test
Unreal and TouchDesigner connectivity test within environment
This approach surfaced issues of latency and instability. Delayed responses also disrupted the sense of intimacy I was aiming for.
To avoid rigid scaling animations, I developed materials that responded dynamically through world displacement driven by noise. This allowed geometry to feel more ‘alive’ as it reacted to sound. However, the technical instability again raised questions about control, leading me to consider more intentional input methods.
Geometry manipulation experiements
Installation and Staging
To ground the virtual experience physically, I designed a raised platform using industrial crates fitted with fluorescent bulbs. The lighting system used a parallel circuit with four bulbs, intentionally diffused by leaving white fabric intact to soften light leakage. This platform became a liminal threshold between the physical and virtual, reinforcing the act of stepping into a space of attention and listening.
Evolution of the platform
Pivotal Changes
At this stage, the project shifted from audio-reactive visuals to a sort of story container. Spoken recordings were attached to floating spheres, with each sphere holding a single narrative rather than contributing to a shared soundscape.
This shift required a restructuring of the system logic. Spheres and audio files were spawned into parallel arrays and assigned shared identifiers to ensure that each story remained correctly paired with its corresponding object. Audio was imported during runtime using a dedicated plugin, and additional safeguards were introduced to prevent existing recordings from being overwritten.
Spawning story containers while synchronously generating audio files
Initially, recording was triggered automatically through voice detection. However, this approach quickly proved ineffective, as natural pauses in speech caused recordings to fragment into multiple unintended clips. To address this, the interaction was redesigned to require participants to intentionally press a key to begin recording.
Spawning using voice
Spawning while pressing and holding a button
To ensure stories were encountered individually, audio only played when a participant approached a sphere. Visual feedback of this interaction was provided via Unreal’s Niagara particle system, which triggered upon contact. This reinforced the sense of an intimate, embodied exchange rather than a passive listening experience.
Participant interaction with a sphere
From Virtual Camera to VR
After initial tests with a virtual camera and a mobile-based approach, ultimately VR emerged as the most coherent solution, preserving immersion while integrating spatial navigation and audio listening.
The transition to VR introduced complex technical hurdles, specifically in developing an asymmetrical multiplayer design. One participant acts as the VR listener, one as a PC-based recorder, as well as the implementation of one non-traditional third-person VR perspective. Despite these challenges, I felt the depth of the final experience justified the technical steepness.
Participants' POV
Central Screen's POV
Reflection
Sonder did not emerge from a fixed technical plan, but from a consistent emotional intention. From the outset, I was less concerned with locking down a final system than with allowing room for experimentation, particularly as this was my first time working with Unreal Engine.
Although the system could theoretically exist as a single application, I deliberately separated it into distinct stations that asked participants to take different actions. The separation of stations becomes less a technical necessity and more a conceptual one. Distributing actions across multiple stations prevents participants from performing every gesture in one place and immediately ending the experience, reinforcing the idea that listening, speaking, and observing require time and attention. If I had more time, I would have explored additional variations of the virtual environment, with how and where the sphere move. But overall, I feel the project achieved what it set out to do.